### Import statewide pedestrian crash data 2017-2019 data and preview
pedcrashes<-read_csv("../Data/pedcrashes20172019.csv")
## Rows: 6079 Columns: 59
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (48): crash_numb, city_town_name, crash_date, crash_severity_descr, cra...
## dbl (10): year, numb_vehc, district_num, numb_fatal_injr, numb_nonfatal_inj...
## time (1): crash_time_2
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
paged_table((pedcrashes))
## Warning: One or more parsing issues, see `problems()` for details
### Select fields of interest and preview
pedcrashes<-pedcrashes%>%
select(crash_numb, max_injr_svrty_cl, drvr_cntrb_circ_cl, road_surf_cond_descr, road_cntrb_descr, traf_cntrl_devc_type_descr, ambnt_light_descr)
paged_table(head(pedcrashes))
###Summarize fields of interest
summary(pedcrashes)
## crash_numb max_injr_svrty_cl drvr_cntrb_circ_cl road_surf_cond_descr
## Length:6079 Length:6079 Length:6079 Length:6079
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
## road_cntrb_descr traf_cntrl_devc_type_descr ambnt_light_descr
## Length:6079 Length:6079 Length:6079
## Class :character Class :character Class :character
## Mode :character Mode :character Mode :character
### Identify unique values of Max Injury Severity Reported
table(select(pedcrashes, max_injr_svrty_cl))
##
## Fatal injury (K) No Apparent Injury (O)
## 117 236
## No injury Non-fatal injury - Incapacitating
## 867 539
## Non-fatal injury - Non-incapacitating Non-fatal injury - Possible
## 1804 1522
## Possible Injury (C) Suspected Minor Injury (B)
## 265 552
## Suspected Serious Injury (A)
## 174
##Use case_when to create new field "injury_level" and mutate max_injr_svrty_cl into numeric scale using the KABCO injury classification scale
#### K=5, A=4, B=3, C=2, O=1
pedcrashes1<-pedcrashes%>%
mutate(injury_level = case_when(
max_injr_svrty_cl == "Fatal injury (K)" ~ 5,
max_injr_svrty_cl == "Non-fatal injury - Incapacitating"~ 4,
max_injr_svrty_cl == "Suspected Serious Injury (A)" ~ 4,
max_injr_svrty_cl == "Non-fatal injury - Non-incapacitating" ~ 3,
max_injr_svrty_cl == "Suspected Minor Injury (B)" ~ 3,
max_injr_svrty_cl == "Non-fatal injury - Possible" ~ 2,
max_injr_svrty_cl == "Possible Injury (C)" ~ 2,
max_injr_svrty_cl == "No injury" ~ 1,
max_injr_svrty_cl == "No Apparent Injury (O)" ~ 1,
TRUE ~ -1))
pedcrashesgraph<-pedcrashes1%>%
filter(injury_level != -1) %>%
mutate(injury_label = case_when(
injury_level == 5 ~ "Fatal injury (K)",
injury_level == 4 ~ "Suspected Serious Injury (A)",
injury_level == 3 ~ "Possible Injury (C)",
injury_level == 2 ~ "Suspected Minor Injury (B)",
injury_level == 1 ~ "No Apparent Injury (O)",
TRUE ~ "NA"))
### Create bar chart for crash severity distribution
barchart<-pedcrashesgraph%>%
ggplot(aes(injury_label)) +geom_bar(stat="count",width=.7, fill="lightblue") +
theme(axis.text.x = element_text(angle = 10, size= 13, vjust= 1, hjust=1))+ ### adjust x-axis label orientation
geom_text(stat='count', aes(label=..count..), size= 6, vjust=1.6 )+ ### add count labels
ggtitle("Massachusetts 2017-2019 Pedestrian Crashes by Crash Severity")+
xlab("Crash Severity")+
ylab("Number of Crashes")+
theme(plot.title = element_text(size=22, face="bold"),
axis.text=element_text(size=12),
axis.title=element_text(size=14,face="bold"))
barchart
### Export bar chart
ggsave("pedcrashbarchart.png", plot=barchart,width = 330,
height = 200,
units = "mm")
##Use case_when to create new field "is_ka" and mutate max_injr_svrty_cl into dummy for 1= KA, 0 = not a KA crash
pedcrashes2<-pedcrashes1%>%
mutate(is_ka = case_when(
max_injr_svrty_cl == "Fatal injury (K)" ~ 1,
max_injr_svrty_cl == "Non-fatal injury - Incapacitating"~ 1,
max_injr_svrty_cl == "Suspected Serious Injury (A)" ~ 1,
max_injr_svrty_cl == "Non-fatal injury - Non-incapacitating" ~ 0,
max_injr_svrty_cl == "Suspected Minor Injury (B)" ~ 0,
max_injr_svrty_cl == "Non-fatal injury - Possible" ~ 0,
max_injr_svrty_cl == "Possible Injury (C)" ~ 0,
max_injr_svrty_cl == "No injury" ~ 0,
max_injr_svrty_cl == "No Apparent Injury (O)" ~ 0,
TRUE ~ -1))
table(select(pedcrashes2, is_ka))
##
## -1 0 1
## 3 5246 830
### Identify unique values of Road Surface Condition description
table(select(pedcrashes, road_surf_cond_descr))
##
## Dry Ice
## 4925 29
## Sand, mud, dirt, oil, gravel Slush
## 12 11
## Snow Water (standing, moving)
## 53 1
## Wet
## 1045
### create dummy road_surface where 1 = road surface altered, 0 = dry road surface
pedcrashes3<-pedcrashes2%>%
mutate(road_surface = case_when(
road_surf_cond_descr == "Dry" ~ 0,
road_surf_cond_descr == "Ice"~ 1,
road_surf_cond_descr == "Sand, mud, dirt, oil, gravel"~ 1,
road_surf_cond_descr == "Slush" ~ 1,
road_surf_cond_descr == "Snow" ~ 1,
road_surf_cond_descr == "Water (standing, moving)" ~ 1,
road_surf_cond_descr == "Wet" ~ 1,
road_cntrb_descr == "Road surface condition (wet, icy, snow, slush, etc.)" ~ 1,
TRUE ~ -1))
table(select(pedcrashes3, road_surface))
##
## -1 0 1
## 3 4925 1151
### Identify unique values of Road Contributing Circumstance Descriptions
table(select(pedcrashes, road_cntrb_descr))
##
## Debris
## 3
## None
## 5565
## Obstruction in roadway
## 17
## Other
## 79
## Road surface condition (wet, icy, snow, slush, etc.)
## 202
## Rut, holes, bumps
## 6
## Shoulders (none, low, soft)
## 4
## Toll/booth/plaza related
## 2
## Traffic congestion related
## 148
## Traffic control device inoperative, missing, or obscured
## 13
## Work zone (construction/maintenance/utility)
## 32
## Worn, travel-polished surface
## 5
### create dummy road_circumstances where 1 = None , 0 = road contributing circumstance
pedcrashes4<-pedcrashes3%>%
mutate(road_circumstances = case_when(
road_cntrb_descr == "None" ~ 0,
road_cntrb_descr == "Debris"~ 1,
road_cntrb_descr == "Obstruction in roadway"~ 1,
road_cntrb_descr == "Other" ~ 1,
road_cntrb_descr == "Road surface condition (wet, icy, snow, slush, etc.)" ~ 0, ###coded 0 since value accounted for in road_surface
road_cntrb_descr == "Rut, holes, bumps" ~ 1,
road_cntrb_descr == "Shoulders (none, low, soft)" ~ 1,
road_cntrb_descr == "Toll/booth/plaza related" ~ 1,
road_cntrb_descr == "Traffic congestion related" ~ 1,
road_cntrb_descr == "Traffic control device inoperative, missing, or obscured" ~ 1,
road_cntrb_descr == "Work zone (construction/maintenance/utility)" ~ 1,
road_cntrb_descr == "Worn, travel-polished surface" ~ 1,
TRUE ~ -1))
table(select(pedcrashes4, road_circumstances))
##
## -1 0 1
## 3 5767 309
### Identify unique values of Ambient Light Descriptions
table(select(pedcrashes,ambnt_light_descr))
##
## Dark - lighted roadway Dark - roadway not lighted
## 1402 270
## Dark - unknown roadway lighting Dawn
## 40 71
## Daylight Dusk
## 4130 163
### create dummy ambient_light where 1 = Daylight, 0 = Dark or potential lack of light
pedcrashes5<-pedcrashes4%>%
mutate(ambient_light = case_when(
ambnt_light_descr == "Dark - lighted roadway" ~ 0,
ambnt_light_descr == "Dark - unknown roadway lighting"~ 0,
ambnt_light_descr == "Dark - roadway not lighted" ~ 0,
ambnt_light_descr == "Dawn" ~ 0,
ambnt_light_descr == "Daylight" ~ 1,
ambnt_light_descr == "Dusk" ~ 0,
TRUE ~ -1))
table(select(pedcrashes5, ambient_light))
##
## -1 0 1
## 3 1946 4130
### Identify unique values of Road Contributing Circumstance Descriptions
table(select(pedcrashes, traf_cntrl_devc_type_descr))
##
## Flashing traffic control signal No controls
## 65 3807
## Railway crossing device School zone signs
## 2 29
## Stop signs Traffic control signal
## 856 1162
## Warning signs Yield signs
## 79 76
### create dummy traffic_control_device where 1 = no traffic control device present, 0 = no traffic controls present
pedcrashes6<-pedcrashes5%>%
mutate(traffic_control_device = case_when(
traf_cntrl_devc_type_descr == "No controls" ~ 0,
traf_cntrl_devc_type_descr == "Yield signs"~ 1,
traf_cntrl_devc_type_descr == "Traffic control signal"~ 1,
traf_cntrl_devc_type_descr == "Flashing traffic control signal" ~ 1,
traf_cntrl_devc_type_descr == "Warning signs" ~ 1,
traf_cntrl_devc_type_descr == "Railway crossing device" ~ 1,
traf_cntrl_devc_type_descr == "Yield signs" ~ 1,
traf_cntrl_devc_type_descr == "School zone signs" ~ 1,
traf_cntrl_devc_type_descr == "Stop signs" ~ 1,
TRUE ~ -1))
table(select(pedcrashes6, traffic_control_device))
##
## -1 0 1
## 3 3807 2269
### Identify unique values of Driver Contributing Circumstance Descriptions
## Many values, remove comment to run: table(select(pedcrashes, drvr_cntrb_circ_cl))
### create dummy traffic_control_device where 1 = at least 1 driver had improper driving, 0 = no improper driving for at least 1 driver, others may be unknown
pedcrashes7<-pedcrashes6%>%
mutate(improper_driving = case_when(
str_detect(drvr_cntrb_circ_cl, "speed limit") ~ 1,
str_detect(drvr_cntrb_circ_cl, "yield") ~ 1,
str_detect(drvr_cntrb_circ_cl, "too closely") ~ 1,
str_detect(drvr_cntrb_circ_cl, "improper turn") ~ 1,
str_detect(drvr_cntrb_circ_cl, "visibility") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Visibility") ~ 1,
str_detect(drvr_cntrb_circ_cl, "too fast") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Distracted") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Inattention") ~ 1,
str_detect(drvr_cntrb_circ_cl, "reckless") ~ 1,
str_detect(drvr_cntrb_circ_cl, "aggressive") ~ 1,
str_detect(drvr_cntrb_circ_cl, "steering") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Disregarded traffic signs") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Other improper action") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Swerving") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Glare") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Failure") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Illness") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Fatigue") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Emotional") ~ 1,
str_detect(drvr_cntrb_circ_cl, "History heart/epilepsy/fainting") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Wrong side or wrong way") ~ 1,
str_detect(drvr_cntrb_circ_cl, "Physical impairment") ~ 1,
TRUE ~ 0))
table(select(pedcrashes7, improper_driving))
##
## 0 1
## 3134 2945
### Check if all driver contributing circumstances are captured, excluding defective equipment
filtercheck<-pedcrashes7%>%
filter(improper_driving == 0) %>%
paged_table()
table(select(filtercheck, drvr_cntrb_circ_cl))
##
## D1: (),(No improper driving)
## 7
## D1: (No improper driving)
## 2937
## D1: (No improper driving) / D2: (No improper driving)
## 28
## D1: (No improper driving) / D2: (No improper driving) / D3: (No improper driving)
## 2
## D1: (No improper driving) / D2: (No improper driving) / D3: (No improper driving) / D4: (No improper driving)
## 1
## D1: (No improper driving) / D2: (No improper driving),(No improper driving)
## 1
## D1: (No improper driving) / D2: (Unknown)
## 3
## D1: (No improper driving) / D4: (No improper driving)
## 1
## D1: (No improper driving),(No improper driving)
## 120
## D1: (No improper driving),(Unknown)
## 9
## D1: (Operating defective equipment)
## 2
## D1: (Unknown) / D2: (No improper driving)
## 5
## D1: (Unknown) / D2: (No improper driving) / D3: (No improper driving)
## 1
## D1: (Unknown) / D3: (No improper driving)
## 1
## D1: (Unknown) / D3: (No improper driving) / D4: (Unknown)
## 1
## D1: (Unknown),(No improper driving)
## 5
## D1: (Unknown),(Unknown) / D2: (No improper driving),(No improper driving)
## 1
## D2: (No improper driving)
## 4
## D2: (No improper driving) / D3: (No improper driving)
## 1
## D3: (No improper driving),(No improper driving) / D4: (No improper driving),(No improper driving)
## 1
##Filter to only necessary fields
###Filter out unknowns for all variables and view data with relevant fields
pedcrashesclean<-pedcrashes7%>%
select(injury_level, is_ka, road_surface, road_circumstances, ambient_light, traffic_control_device, improper_driving)%>%
filter(traffic_control_device!= -1, na.rm= TRUE)
summary(pedcrashesclean)
## injury_level is_ka road_surface road_circumstances
## Min. :1.000 Min. :0.0000 Min. :0.0000 Min. :0.00000
## 1st Qu.:2.000 1st Qu.:0.0000 1st Qu.:0.0000 1st Qu.:0.00000
## Median :3.000 Median :0.0000 Median :0.0000 Median :0.00000
## Mean :2.499 Mean :0.1366 Mean :0.1894 Mean :0.05086
## 3rd Qu.:3.000 3rd Qu.:0.0000 3rd Qu.:0.0000 3rd Qu.:0.00000
## Max. :5.000 Max. :1.0000 Max. :1.0000 Max. :1.00000
## ambient_light traffic_control_device improper_driving
## Min. :0.0000 Min. :0.0000 Min. :0.0000
## 1st Qu.:0.0000 1st Qu.:0.0000 1st Qu.:0.0000
## Median :1.0000 Median :0.0000 Median :0.0000
## Mean :0.6797 Mean :0.3734 Mean :0.4847
## 3rd Qu.:1.0000 3rd Qu.:1.0000 3rd Qu.:1.0000
## Max. :1.0000 Max. :1.0000 Max. :1.0000
###Create Training data
set.seed(1)
trainingsetsize<-pedcrashesclean%>%
nrow()%>%
multiply_by(0.7)%>%
round()
train_indices<-sample(1:nrow(pedcrashesclean), trainingsetsize)
train<-pedcrashesclean[train_indices,]
test<-pedcrashesclean[-train_indices,]
ols_levels<-lm(injury_level ~ .-is_ka, data = train)
ols_binary<-lm(is_ka ~ .-injury_level, data = train)
par(mfrow = c(2,3))
plot(ols_binary, which = 1:6)
par(mfrow = c(2,3))
plot(ols_levels, which = 1:6)
# backward elimination
step(object = ols_levels, # specify full model
direction = "backward") # specify backward
## Start: AIC=-247.45
## injury_level ~ (is_ka + road_surface + road_circumstances + ambient_light +
## traffic_control_device + improper_driving) - is_ka
##
## Df Sum of Sq RSS AIC
## - road_surface 1 1.1283 4002.4 -248.25
## - road_circumstances 1 1.3730 4002.7 -247.99
## <none> 4001.3 -247.45
## - traffic_control_device 1 9.2060 4010.5 -239.68
## - improper_driving 1 24.6577 4026.0 -223.32
## - ambient_light 1 25.5457 4026.9 -222.38
##
## Step: AIC=-248.25
## injury_level ~ road_circumstances + ambient_light + traffic_control_device +
## improper_driving
##
## Df Sum of Sq RSS AIC
## - road_circumstances 1 1.3532 4003.8 -248.81
## <none> 4002.4 -248.25
## - traffic_control_device 1 9.3477 4011.8 -240.33
## - ambient_light 1 24.4184 4026.9 -224.38
## - improper_driving 1 24.4951 4026.9 -224.30
##
## Step: AIC=-248.81
## injury_level ~ ambient_light + traffic_control_device + improper_driving
##
## Df Sum of Sq RSS AIC
## <none> 4003.8 -248.81
## - traffic_control_device 1 9.4998 4013.3 -240.74
## - improper_driving 1 24.6447 4028.4 -224.72
## - ambient_light 1 24.8112 4028.6 -224.54
##
## Call:
## lm(formula = injury_level ~ ambient_light + traffic_control_device +
## improper_driving, data = train)
##
## Coefficients:
## (Intercept) ambient_light traffic_control_device
## 2.58184 -0.16408 -0.09745
## improper_driving
## 0.15285
###test backwards elimination model
br_fit<-lm(formula = injury_level ~ ambient_light + traffic_control_device + improper_driving, data = train)
summary(br_fit)
##
## Call:
## lm(formula = injury_level ~ ambient_light + traffic_control_device +
## improper_driving, data = train)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.7347 -0.5818 0.2653 0.5822 2.5822
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.58184 0.03118 82.797 < 2e-16 ***
## ambient_light -0.16408 0.03198 -5.131 3.01e-07 ***
## traffic_control_device -0.09745 0.03069 -3.175 0.00151 **
## improper_driving 0.15285 0.02989 5.114 3.29e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.9707 on 4249 degrees of freedom
## Multiple R-squared: 0.0136, Adjusted R-squared: 0.01291
## F-statistic: 19.53 on 3 and 4249 DF, p-value: 1.422e-12
###Create and print diagnostics plots for backwards regression model
png("diagnosticsplots.png", width=15, height=6, units='in', res=300)
par(mfrow = c(1:2))
plot(br_fit, which=1:2, ask=FALSE)
dev.off()
## quartz_off_screen
## 2
###Model Evaluation- run models using testing data, generate AIC, BIC, PRESS, MSE, RSME
predict(br_fit, test) %>% subtract(test$injury_level) %>%
raise_to_power(2) %>% mean() -> MSE_br_fit
predict(ols_binary, test) %>% subtract(test$is_ka) %>%
raise_to_power(2) %>% mean() -> MSE_ols_binary
predict(ols_levels, test) %>% subtract(test$injury_level) %>%
raise_to_power(2) %>% mean() -> MSE_ols_levels
Press1<-press(br_fit)
BIC1<-BIC(br_fit)
AIC1<-AIC(br_fit)
Press2<-press(ols_levels)
BIC2<-BIC(ols_levels)
AIC2<-AIC(ols_levels)
###visualize metrics in table
cat(' ', " MSE", ' RMSE', ' PRESS',' AIC', ' BIC', '\n',
'Backwards Elim:', MSE_br_fit, sqrt(MSE_br_fit), Press1, AIC1, BIC1, '\n',
'All Predictors:', MSE_ols_levels, sqrt(MSE_ols_levels), Press2, AIC2,BIC2, '\n')
## MSE RMSE PRESS AIC BIC
## Backwards Elim: 0.9702171 0.984996 4011.383 11822.68 11854.45
## All Predictors: 0.9714831 0.9856384 4012.704 11824.04 11868.53
stargazer(ols_levels, br_fit, title="Table 1. Regression Results",
dep.var.labels=c("Injury Severity Level"),
covariate.labels=c("Pavement Friction Impaired", "Road Contributing Circumstance",
"Ambient Light","Traffic Control Device","Improper Driving"),
column.labels=c("All Predictors","Backwards Elimination"),
add.lines=list(c("AIC", "11824.04", "11822.68"), c("BIC","11868.53","11854.45"), c("PRESS","4012.704","4011.383")),
type="text", out="stargazertable.txt", ci=TRUE, ci.level=0.05, single.row=TRUE)
##
## Table 1. Regression Results
## ====================================================================================
## Dependent variable:
## -----------------------------------------------------
## Injury Severity Level
## All Predictors Backwards Elimination
## (1) (2)
## ------------------------------------------------------------------------------------
## Pavement Friction Impaired -0.042 (-0.045, -0.040)
## Road Contributing Circumstance -0.081 (-0.085, -0.077)
## Ambient Light -0.171*** (-0.173, -0.169) -0.164*** (-0.166, -0.162)
## Traffic Control Device -0.096*** (-0.098, -0.094) -0.097*** (-0.099, -0.096)
## Improper Driving 0.153*** (0.151, 0.155) 0.153*** (0.151, 0.155)
## Constant 2.598*** (2.596, 2.600) 2.582*** (2.580, 2.584)
## ------------------------------------------------------------------------------------
## AIC 11824.04 11822.68
## BIC 11868.53 11854.45
## PRESS 4012.704 4011.383
## Observations 4,253 4,253
## R2 0.014 0.014
## Adjusted R2 0.013 0.013
## Residual Std. Error 0.971 (df = 4247) 0.971 (df = 4249)
## F Statistic 12.248*** (df = 5; 4247) 19.532*** (df = 3; 4249)
## ====================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
###Export stargazer as html too
stargazer(ols_levels, br_fit, title="Table 1. Regression Results",
dep.var.labels=c("Injury Severity Level"),
covariate.labels=c("Pavement Friction Impaired", "Road Contributing Circumstance",
"Ambient Light","Traffic Control Device","Improper Driving"),
column.labels=c("All Predictors","Backwards Elimination"),
add.lines=list(c("AIC", "11824.04", "11822.68"), c("BIC","11868.53","11854.45"), c("PRESS","4012.704","4011.383")),
type="html", out="stargazertable.html", ci=TRUE, ci.level=0.05, single.row=TRUE)
##
## <table style="text-align:center"><caption><strong>Table 1. Regression Results</strong></caption>
## <tr><td colspan="3" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left"></td><td colspan="2"><em>Dependent variable:</em></td></tr>
## <tr><td></td><td colspan="2" style="border-bottom: 1px solid black"></td></tr>
## <tr><td style="text-align:left"></td><td colspan="2">Injury Severity Level</td></tr>
## <tr><td style="text-align:left"></td><td>All Predictors</td><td>Backwards Elimination</td></tr>
## <tr><td style="text-align:left"></td><td>(1)</td><td>(2)</td></tr>
## <tr><td colspan="3" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left">Pavement Friction Impaired</td><td>-0.042 (-0.045, -0.040)</td><td></td></tr>
## <tr><td style="text-align:left">Road Contributing Circumstance</td><td>-0.081 (-0.085, -0.077)</td><td></td></tr>
## <tr><td style="text-align:left">Ambient Light</td><td>-0.171<sup>***</sup> (-0.173, -0.169)</td><td>-0.164<sup>***</sup> (-0.166, -0.162)</td></tr>
## <tr><td style="text-align:left">Traffic Control Device</td><td>-0.096<sup>***</sup> (-0.098, -0.094)</td><td>-0.097<sup>***</sup> (-0.099, -0.096)</td></tr>
## <tr><td style="text-align:left">Improper Driving</td><td>0.153<sup>***</sup> (0.151, 0.155)</td><td>0.153<sup>***</sup> (0.151, 0.155)</td></tr>
## <tr><td style="text-align:left">Constant</td><td>2.598<sup>***</sup> (2.596, 2.600)</td><td>2.582<sup>***</sup> (2.580, 2.584)</td></tr>
## <tr><td colspan="3" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left">AIC</td><td>11824.04</td><td>11822.68</td></tr>
## <tr><td style="text-align:left">BIC</td><td>11868.53</td><td>11854.45</td></tr>
## <tr><td style="text-align:left">PRESS</td><td>4012.704</td><td>4011.383</td></tr>
## <tr><td style="text-align:left">Observations</td><td>4,253</td><td>4,253</td></tr>
## <tr><td style="text-align:left">R<sup>2</sup></td><td>0.014</td><td>0.014</td></tr>
## <tr><td style="text-align:left">Adjusted R<sup>2</sup></td><td>0.013</td><td>0.013</td></tr>
## <tr><td style="text-align:left">Residual Std. Error</td><td>0.971 (df = 4247)</td><td>0.971 (df = 4249)</td></tr>
## <tr><td style="text-align:left">F Statistic</td><td>12.248<sup>***</sup> (df = 5; 4247)</td><td>19.532<sup>***</sup> (df = 3; 4249)</td></tr>
## <tr><td colspan="3" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left"><em>Note:</em></td><td colspan="2" style="text-align:right"><sup>*</sup>p<0.1; <sup>**</sup>p<0.05; <sup>***</sup>p<0.01</td></tr>
## </table>